SSSAJ Grow Your Career with SSSA
HOME HELP FEEDBACK SUBSCRIPTIONS ARCHIVE SEARCH TABLE OF CONTENTS
QUICK SEARCH:   [advanced]


     

Published online 2 June 2005
Published in Soil Sci Soc Am J 69:1016-1025 (2005)
DOI: 10.2136/sssaj2003.0093
© 2005 Soil Science Society of America
677 S. Segoe Rd., Madison, WI 53711 USA
This Article
Right arrow Abstract Freely available
Right arrow Figures Only
Right arrow Full Text (PDF) Free
Right arrow Alert me when this article is cited
Right arrow Alert me if a correction is posted
Services
Right arrow Similar articles in this journal
Right arrow Similar articles in ISI Web of Science
Right arrow Alert me to new issues of the journal
Right arrow Download to citation manager
Right arrow reprints & permissions
Citing Articles
Right arrow Citing Articles via HighWire
Right arrow Citing Articles via Google Scholar
Google Scholar
Right arrow Articles by Fang, F.
Right arrow Articles by Hatch, L. K.
Right arrow Search for Related Content
PubMed
Right arrow Articles by Fang, F.
Right arrow Articles by Hatch, L. K.
GeoRef
Right arrow GeoRef Citation
Agricola
Right arrow Articles by Fang, F.
Right arrow Articles by Hatch, L. K.
Related Collections
Right arrow Water Quality
Right arrow Nutrient Management
Right arrow Soil Chemistry

Soil Chemistry

Characterization of Soil Algal Bioavailable Phosphorus in the Minnesota River Basin

F. Fanga,e, P. L. Brezonikb,g,*, D. J. Mullac and L. K. Hatchd,f

a Water Resources Science Graduate Program, 500 Pillsbury Dr. SE, Univ. of Minnesota, Minneapolis, MN 55455
b Dep. of Civil Engineering, Univ. of Minnesota, Minneapolis, MN 55455
c Dep. of Soil, Water and Climate, Univ. of Minnesota, St. Paul, MN 55108
d Water Resources Center, Univ. of Minnesota, St. Paul, MN 55108
e Current address: Kieser & Associates–Environmental Science and Engineering, 536 E. Michigan Ave., Suite 300, Kalamazoo, MI 49007
f Current address: Minnehaha Creek Watershed District, Deephaven, MN 55391
g BES-565, National Science Foundation, Arlington, VA 22230

* Corresponding author (brezonik{at}umn.edu)


   ABSTRACT
TOP
 ABSTRACT
INTRODUCTION
 MATERIALS AND METHODS
 RESULTS AND DISCUSSION
 CONCLUSIONS
 REFERENCES
 
Soil algal bioavailable P (ABP) is the P fraction that contributes most directly to eutrophication in freshwaters affected by agricultural nonpoint-source pollution. There are uncertainties regarding the algal bioavailability of P sorbed to calcareous glacial till soils in the upper Midwest. The ABP of soil samples with a broad range of pH and calcium carbonate content from six sites across the Minnesota River Basin (MRB) was measured by algal bioassay, and relationships with various soil physical and chemical properties were studied. For soils of the MRB, a major agricultural watershed in the upper Midwest, ABP was significantly correlated (p < 0.001) with Bray P, Mehlich-III P, NaOH P, Oxalate P, and Fe-paper P. Among them, Fe-paper P approximated ABP best, particularly for calcareous soils. For acidic soils, amorphous Fe and Al apparently were the primary P retention agents in soil particles. Soil sorption data were well described by the linearized Langmuir sorption model. Although the P sorption maximum ({Gamma}{infty}) could not be predicted from basic soil physical and chemical properties, the sorption energy constant (b) was highly correlated with soil pH, clay content, and organic matter (OM) content. A P saturation index (PSIs) that uses sorptivity ({Gamma}{infty} x b) as the measure of sorption capacity gave the best estimate of soil ABP among the predictors used in this study. Phosphorus saturated index itself can be approximated by the widely available Bray-P value for soils in the MRB.

Abbreviations: ABP, algal bioavailable phosphorus • CCE, calcium carbonate equivalent • EPC, equilibrium phosphorus concentration • MPCA, Minnesota Pollution Control Agency • MRB, Minnesota River Basin • OM, organic matter • PSIm, phosphorus saturation index based on sorption maximum • PSIs, phosphorus saturation index based on sorptivity • SRP, soluble reactive phosphorus • STP, soil test phosphorus • TP, total phosphorus


   INTRODUCTION
TOP
 ABSTRACT
 INTRODUCTION
MATERIALS AND METHODS
 RESULTS AND DISCUSSION
 CONCLUSIONS
 REFERENCES
 
THE MOST RECENT U.S. national water quality inventory showed that nonpoint-source pollution from agriculture is the leading source of water quality problems in 48% of the rivers and streams and 41% of the lakes with impaired water quality in the nation (USEPA, 2002). In terms of pollutants, nutrients were found to be the leading stressor in over 20% of the rivers and streams and 50% of the lakes. Phosphorus has been widely recognized as the limiting nutrient for the eutrophication process in fresh water bodies. Both state and federal regulatory agencies are pursuing policies to reduce P losses from agricultural sources, and a variety of best management practices (BMPs) have been proposed for this purpose.

Because of the strong retention of P by soil particles, reducing the loss of soil to runoff is an important step in reducing P pollution. Phosphorus exists in soil in different forms, not all of which are readily available for the excessive algal growth associated with eutrophication. Therefore, it is crucial to quantify the ABP of the soils in a watershed when evaluating the impact of agricultural nonpoint sources on water quality and comparing the cost-effectiveness of control measures for the watershed.

Many factors affect the amount of ABP of a particular soil. Farming practices, such as tillage style, source of fertilizer, and application method and rate, are critical management factors in determining the overall amount of ABP in a soil. However, soil physical and chemical properties, such as pH, OM content, clay content, and mineral composition, influence the extent and intensity of soil P sorption–desorption and the forms of P retained by a soil (e.g., Beek et al., 1980; Ryan et al., 1985). In turn, sorption–desorption processes affect the quantity and timing of the retained P available for algal growth.

The traditional and most direct method to measure soil ABP involves algal bioassays. This process can take several months and involves expertise and equipment not readily available to farmers and pollution control workers (Sharpley et al., 1991; Fang et al., 2002). Consequently, common soil test methods used to determine crop-available P (soil test P or STP), such as Bray P and Mehlich-I P, have been used as surrogates of soil ABP (Sims, 1993). Other methods, such as NaOH extraction (Dorich et al., 1985; Sharpley et al., 1991) and the iron oxide-impregnated filter-paper technique were developed to better approximate soil ABP (Sharpley, 1993; Sharpley et al., 1994; Chardon et al., 1996), but most studies using these methods have been done on acidic soils in the southern and eastern USA and western Europe.

Alternatively, knowledge of soil characteristics can be used to develop predictive relationships for soil ABP (Sharpley, 1993; Sharpley et al., 1994; Chardon et al., 1996). For example, the P sorption–desorption capacity of a soil may provide useful information on the bioavailability of soil P. However, measuring this characteristic is a lengthy laboratory process that requires expertise (Nair et al., 1984). If soil sorption–desorption capacity could be estimated from basic physical and chemical properties (e.g., clay content, OM content, and Fe and Al content), its use to estimate soil ABP would be facilitated. To our knowledge, no studies have evaluated the relationships between P sorption and bioavailable P in calcareous soils. Although P sorption in calcareous soils has been the subject of a handful of studies (Ryan et al., 1985; Sharpley and Smith, 1985; Solis and Torrent, 1989; Holford and Mattingly, 1974; Zhou and Li, 2001), the primary focus of these studies has been on the relative importance of iron oxide versus calcium carbonate in controlling P sorption capacities and concentrations of labile P. The applicability of relationships and methods for estimating ABP in acidic soils to calcareous soils in upper Midwest, where many soils were formed from calcareous glacial till, is uncertain.

The study area of this research, the MRB is a major agricultural watershed in southern Minnesota (Fig. 1) . It has suffered severe P pollution problems for decades. Nonpoint sources, primarily surface and subsurface runoff from agricultural fields, account for approximately 74% of the river's P loadings (MPCA, 2001). The MRB has three basic types of soil parent material: moraines, glacial till plains, and glacial lake beds. Although they mostly originated from calcareous glacial material, a diverse group of soils, ranging from calcareous soils in the west to acidic soils in the east, has developed in the MRB as a result of different local natural conditions (e.g., drainage, landscape, and precipitation) and agricultural practices (e.g., row crops versus pasture). Although calcareous soils are the focus of this study, comparison of soils with different pH from the same basin can provide additional insights on the bioavailability of soil-bound P.



View larger version (24K):
[in this window]
[in a new window]
  		Fig. 1. The Minnesota River Basin and the sampling sites.

 
The objectives of this study thus were threefold: (i) to determine the relationships between various STP measurements and ABP for calcareous and other soils of the MRB; (ii) to determine the relationship between soil physical–chemical characteristics and P sorption capacity for soils of the basin; and (iii) to determine the relationship between ABP and P sorption capacity for soils of the basin.


   MATERIALS AND METHODS
TOP
 ABSTRACT
 INTRODUCTION
 MATERIALS AND METHODS
RESULTS AND DISCUSSION
 CONCLUSIONS
 REFERENCES
 
Sampling Sites
The MRB (Fig. 1) in central and southwestern Minnesota has 12 major watersheds and 1208 minor ones in four physiographic regions (Minnesota Pollution Control Agency [MPCA], 1997; Mulla and Mallawatantri, 1997). Mean annual precipitation ranges from 787 mm in the eastern part of the basin to 584 mm in the west. We collected 54 surface soils in July and August 1998 from six sites across the MRB (Fig. 1) selected to represent a wide range of soil characteristics, such as total P (TP), pH, OM content, and calcium carbonate content (Tables 1 and 2). Three of the sites, Morris, Lamberton, and Waseca, are located on University of Minnesota agricultural experiment station lands. Soil characteristics of these stations in total are representative of approximately 50% of croplands in the basin. All soils were collected from land cultivated for corn and soybeans except for the Ivanhoe site, which is a sheep pasture. Management history did not include manure applications. Except for the pasture, all sites were row cropped at the time of sampling.


View this table:
[in this window]
[in a new window]
  		Table 1. Soil characteristics for the 25 samples used for algal bioavailable P (ABP) and P sorption studies.

 

View this table:
[in this window]
[in a new window]
  		Table 2. Statistical summary of physical and chemical properties of all soil samples (n = 54) and the 25 samples used in algal bioavailable P (ABP) and P sorption studies.

 
At each site, three transects (coded 1, 2, and 3) were selected for sampling with 20 to 30 m spacing between them depending on field conditions. On each transect, three sampling locations were selected according to landscape position (slope distribution of the field) and coded as up-slope, mid-slope, and low-slope. Thus, nine soil samples were collected at each site. Samples were coded by the name of the nearby town, slope position, and transect number; for example, Waseca Up 1 is from the up-slope of transect 1 of the site near Waseca, MN. At each location, topsoil up to 5 cm deep was collected by shovel after removal of vegetation and crop debris. We assumed no variation in soil bulk density among sites, which seems reasonable given that the soils are shallow and all sites use similar tillage systems. Samples were placed in 1-gallon ZipLoc bags (S.C. Johnson & Son, Racine, WI) and transferred to the laboratory, where they were sieved through a 2-mm steel mesh after air-drying.

Chemical and Statistical Analyses
For all 54 samples, soil pH, OM content (percentage loss on ignition), calcium carbonate equivalent (CCE, %), TP, Mehlich-III P, Mehlich-III Fe, and Bray P were measured by the Research Analytical Laboratory, University of Minnesota, St. Paul. Bray and Olson extractable P measurements of plant available P are common in Minnesota, with Bray P being used in neutral to acidic soils, and Olson P being used in basic soils. Mehlich-III P is not typically measured by commercial soil testing laboratories in Minnesota. Soil particle-size distribution was analyzed by the standard pipette method (Gee and Bauder, 1996). We selected 24 soils representing typical values and ranges of the eight chemical and physical properties of the 54 samples, to perform the ensuing ABP-related soil characterizations.

For the 24 samples, ammonium oxalate-extractable P (oxalate P) was measured by mixing 30 mL of a solution containing 0.2 M oxalic acid and 0.2 M ammonium oxalate (v:v = 764:1000, pH 3.0 [Pote et al., 1999]) with 0.75 g of soil in 50 mL Nalgene Oak Ridge centrifuge tubes (Nalge Nunc International, Rochester, NY) wrapped in aluminum foil. After shaking end-to-end for 2 h, samples were centrifuged for 25 min at 11 200 x g (12000 rpm), and the supernatant was analyzed for P, Al, and Fe by inductively coupled plasma spectrometer (ICP).

Iron oxide-impregnated filter paper extractable P (Fe-paper P) was measured by a modification (Fang et al., 2002) of the method of Sharpley et al. (1994) and Chardon et al. (1996). Sodium hydroxide P was extracted with a solution of 0.1 M NaOH and 1.0 M NaCl at a soil to solution ratio of 1/500 (wt/v). This method differs from that of Sharpley et al. (1991) in that 1.0 M NaCl was added to minimize P re-adsorption (Barbanti et al., 1994). During the course of the laboratory work, an additional soil sample (St. Peter Mid 3) was analyzed for NaOH P and P sorption–desorption capacity. As a result, there were 25 data points for NaOH P and P sorption–desorption capacity. Soluble reactive P (SRP) extracted from soil samples by the iron oxide-impregnated filter paper and NaOH methods was determined by the ascorbic acid method (Murphy and Riley, 1962; Eaton et al., 1995).

Statistical analyses were performed using the software packages SYSTAT 7.0.1 (SYSTAT Software, 1997, http://www.systat.com) and STATA 8.0 (StatCorp., 2003; http://www.stata.com). Multiple regressions were computed by the forward stepwise procedure, and the most statistically significant regression equations were determined by individual t values of the independent variables and the overall R2 values.

Algal Bioassay
The bioavailability of P from the 24 soil samples was measured by algal bioassays patterned after the method of Sharpley et al. (1991). We used Stephanodiscus hantzschii as the culture organism because it is the most common and abundant diatom in the Minnesota River. Detailed laboratory procedures were presented by Fang et al. (2002). Briefly, batch cultures of S. hantzschii were grown in a full-nutrient medium for 5 wk and then transferred to a P-free medium to induce P starvation. Soil samples were added after 21 d and incubated for another 10 d. Each batch culture was then analyzed for chlorophyll a (chl a). Results for chl a responses were expressed as the ratio of the mean treatment chl a divided by the mean control (no sediment additions) chl a. During data analysis, one sample was found to be an outlier chl a response (based on Cook's distance calculations). Therefore, only 23 chl a data points were used.

Soil Phosphorus Sorption–Desorption
Soil P sorption–desorption was measured by a small modification of the procedure of Nair et al. (1984). Briefly, 0.52 g of soil sample and 13 mL of a P solution (0, 1000, 3000, 5000, 7500, or 10000 µg P L–1 as KH2PO4) were placed into 50-mL Nalgene centrifuge tubes (Nalge Nunc International, Rochester, NY), tumbled in an end-over-end shaker for 24 h, and centrifuged. The supernatant was decanted, and 13 mL of 0.01 M CaCl2 were added to each tube. After shaking again for 24 h, tubes were centrifuged, and SRP in supernatant from each step was measured after appropriate dilution. The amount of entrained P remaining in the centrifuge tubes following the adsorption phase was accounted for by weighing the tubes before and after centrifugation (Fang et al., 2002). The amount of P adsorbed to or desorbed from soil particles was calculated from the SRP data, and "desorbability" was determined as the percentage of P desorbed from samples during the second step. An expanded form of the Langmuir sorption model (Eq. [1]; Pollman, 1983; Fang et al., 2002) was used to interpret the adsorption–desorption data:

[1]
where Co = initial P concentration of solution (µg L–1); x = mass of P released from soil (mg); v = volume of solution (L); m = mass of soil (mg); {Gamma}o = mass of P initially present (mg kg–1 soil); {Gamma}{infty} = maximum sorption (mg kg–1 soil); b = constant related to the sorption energy (L µg–1); and Co + {Delta}x/v = C* = P concentration (µg L–1) in solution phase at equilibrium. Equation [1] is essentially the same as the basic Langmuir equation (C*/{Gamma} = 1/b{Gamma}{infty} + C*/{Gamma}{infty}), where {Gamma} = concentration of P (mg kg–1) on the soil at equilibrium, except that C* and {Gamma} are replaced by experimentally measured quantities. It was used here because it clearly defines some variables in terms of quantities that were measured experimentally. Particularly, it shows {Gamma}o, the initial mass of P on soil particles, indicating its importance in correctly applying the Langmuir model.

Phosphorus Sorption Saturation
The P sorption saturation approach is based on the principle that soil P loss is a function of both the soil P level and the soil P sorption capacity. Breeuwsma et al. (1995) defined the degree of P saturation as the ratio of sorbed P to the P sorption capacity of a soil. Sharpley (1995), Pote et al. (1999), and Fang et al. (2002) applied this concept to runoff studies and found that soil P sorption saturation correlated well with dissolved P in surface runoff. In this study, the correlation of P sorption saturation with soil ABP is explored. A general definition of P sorption saturation can be written as (Sharpley, 1995):

[2]
To apply Eq. [2], we used NaOH P as the measure of extractable P (the numerator) because NaOH P also was used as the initial sorbed P, {Gamma}o, in fitting the Langmuir sorption model. Methods used to estimate {Gamma}o vary widely and include 32P isotopic exchange (Nair et al., 1984) and extraction by 0.1 M NaCl (Syers et al., 1973), 0.5 M NaHCO3 (He et al., 1999; Zhou and Li, 2001), Mehlich I solution (Sallade and Sims, 1997), iron oxide-impregnated filter paper (Chardon and Blaauw, 1998), and acidified ammonium oxalate (van der Zee et al., 1987). Unfortunately, the basis for selecting a method usually is not provided. We used NaOH extraction because it has been applied successfully in both lake sediment and soil P sorption studies (Detenbeck and Brezonik, 1991; Fang et al., 2002) and because of the correlation between NaOH P and ABP in agricultural runoff (Sharpley et al., 1991).

Two measures of sorption capacity were used: (i) sorption maximum, {Gamma}{infty}; and (ii) sorptivity = {Gamma}{infty} x b (He et al., 1999). Results are denoted as PSIm (P saturation index based on sorption maximum; values in percentagaes):

[3]
or PSIs (P saturation index based on sorptivity):

[4]
Equation [4] has units of concentration (mg L–1), and values were on the order of 10–1. For ease of data presentation, we multiplied calculated values of PSIs by 100. Correlations of desorbability, as defined above, and equilibrium phosphorus concentration (EPC) (Brezonik and Pollman, 1999), with soil ABP also were explored.


   RESULTS AND DISCUSSION
TOP
 ABSTRACT
 INTRODUCTION
 MATERIALS AND METHODS
 RESULTS AND DISCUSSION
CONCLUSIONS
 REFERENCES
 
Soil
Table 2 provides summary statistics on physical and chemical properties of all the soil samples and the samples used in P sorption–desorption studies. Among the 54 samples, 19 were acidic (pH < 6.0), 17 calcareous (pH ≥ 7.6), and 18 circumneutral. Soil pH increased from east to west. Most soils with pH ≥ 7.6 were west of Lamberton; most acidic soils were near St. Peter and Waseca. Calcium carbonate equivalent values followed a similar trend; all above-average soils (CCE > 2.2%) were at or west of Lamberton, indicating soils are more calcareous in the drier, western part of the basin. Calcium carbonate equivalent and soil pH are closely related. Below pH 6.8, CCE levels are low and almost constant (approximately 1.4%). At CCE ≥ approximately 5%, pH stabilizes around 8.0 despite a wide range in CCE. The acid buffering capacity of calcareous soils thus is evident.

The content of OM in the soils (Tables 1 and 2) generally was in the medium to high range (Rehm et al., 1995); 26 samples had an OM content > 4.6%, and only eight had < 3.0%. The soils also had generally high clay contents. There was no obvious geographic trend for Mehlich-III Fe and ammonium oxalate extractable Fe and Al (oxalate Fe + Al) except that the soils from Waseca consistently were on the high ends of both variables.

Soil Phosphorus Level
Table 3 summarizes results for seven soil P tests for all soil samples. Bray P is the most common soil P fertility test in Minnesota for soils with pH < 7.4. Olson P (NaHCO3 extraction) is used for soils with pH ≥ 7.4. Because a much stronger base, NaOH, was used as one of the soil P tests in this study, Olson P was not measured for the soils. Fang et al. (2002) showed that NaOH P and Olson P are closely correlated for a subset of soils used in this study. According to Rehm et al. (1995), a Bray P of 21 mg kg–1 is considered very high in Minnesota, and if the Bray P exceeds 25 mg kg–1, no P fertilizer should be applied. Bray-P levels in our samples generally were very high (35 values ≥ 21 mg kg–1 and 31 values ≥ 25 mg kg–1). Among the six sites, the Ivanhoe sheep pasture had the lowest Bray P (<5 mg kg–1).


View this table:
[in this window]
[in a new window]
  		Table 3. Soil test P (STP) results.

 
Soil Algal Bioavailable Phosphorus
All correlations between pairs of STPs were significant at p < 0.001 (r2 = 0.26–0.91; Table 3), indicating substantial overlap of the soil P extracted by the procedures. If we consider chl a response as the "true" indicator of the bioavailability of soil P to algae (ABP), it is apparent that the P extracted by Fe-paper P most closely parallels soil ABP (r2 = 0.68), followed by Mehlich-III P. Oxalate P and NaOH P also were correlated with chl a response (r2 = 0.62). Iron-paper P is recognized as a good approximation of ABP in soil and agricultural runoff (Menon et al., 1990, 1997; Sharpley, 1993). In contrast to the elaborate procedures required to measure chl a response, Fe-paper offers procedural simplicity and flexibility in sample preservation (Sharpley et al., 1994; Chardon et al., 1996). A regression equation was derived to estimate chl a responses of MRB soils from Fe-paper P values:

[5]
The ten calcareous soils (pH ≥ 7.6) with measured Fe-paper P and chl a response had a higher correlation between these measurements than that for all soils (Eq. [5]), and this equation should be used to estimate chl a response for soils with pH ≥ 7.6, such as occur in the western MRB.


[6]

The Iron Oxide-Impregnated Filter Paper Method
The mild conditions used in the Fe-paper procedure resulted in generally lower amounts of extracted P than other STPs yielded (Table 3). Sharpley (1991) obtained similar results in a study of the relationship between Fe-paper P and other STPs on 203 soils from various places in the world. Sharpley concluded that Fe-paper extracted primarily physically bound P and little P from amorphous Al, Fe, or Ca compounds. A few MRB soils with high CCE or very low Bray P (<10 mg kg–1) had slightly higher Fe-paper P than Bray P. This is consistent with Sharpley's finding that more P was extracted by Bray solution than by Fe-paper with increase in the degree of soil weathering. As shown in Table 4, the amount of P extracted by Fe-paper was closely correlated with Bray P for acidic MRB soils but not for calcareous MRB soils. Menon et al. (1997) reported a similar observation. In contrast, Fe-paper P was closely correlated with NaOH P for calcareous soils but not for acidic soils. The strong acid nature of the Bray extractant and strong base nature of NaOH rendered these extractants unsuitable for extracting P from calcareous and acidic soils, respectively. The closeness of the Fe-paper P correlations with the other STPs (Table 3) and with chl a response across all the three pH groups suggests that the Fe-paper method is a good surrogate not only for STP methods intended to estimate plant-available P but also for the algal bioassay intended to estimate ABP in aquatic systems, and the method applies to soils with wide ranges of physical and chemical properties.


View this table:
[in this window]
[in a new window]
  		Table 4. Coefficients of determination (r2) between Fe-paper P (iron oxide-impregnated filter paper extractable P) and other soil test P values (STPs) with the soils divided into three pH groups.

 
An interesting finding from our study is the excellent correlations between Mehlich-III P and Fe-paper P for all soils (r2 = 0.91), including acidic and calcareous soils (r2 = 0.89 and 0.97, respectively; all p < 0.001; Tables 3 and 4). This differs from Sharpley's (1991) results, which showed Mehlich-III P and Fe-paper P were well correlated for noncalcareous soils but not calcareous soils. In addition, regression of Mehlich-III P versus Fe-paper P indicated that the regression slope increased from acidic to calcareous soils, signifying stronger extraction of P by Mehlich-III extractant from calcareous soils examined in this study. The high buffering capacity of the Mehlich-III extractant (Mehlich, 1984) may explain its consistent strong P extraction from soils with very different pH values, but further studies are needed to fully understand this issue.

Soil Phosphorus Sorption
Table 5 summarizes the sorption parameter values obtained from linearized Langmuir model fitting for the 25 soils. There was good agreement between linear and nonlinear fits of the sorption–desorption data to the Langmuir model (Table 6), which indicates that fitting the data to the linearized model did not cause spurious self-correlations. Because linear fits yielded r2 > 0.98 in all cases and because linear models generally are superior to nonlinear ones for estimating parameters, we used parameters from the linearized model in further analysis of the sorption behavior of the soils.


View this table:
[in this window]
[in a new window]
  		Table 5. Langmuir P sorption–desorption results for the 25 Minnesota River Basin soils. Summary of sorption parameters based on linearized Langmuir model.

 

View this table:
[in this window]
[in a new window]
  		Table 6. Coefficients of determination (r2) for parameters from linear and nonlinear fits of Langmuir model.

 
Phosphorus Sorption Maximum ({Gamma}{infty})
The P sorption maximum ({Gamma}{infty}) and sorption energy constant (b) of the Langmuir model ranged from 180 to 335 mg kg–1 and from 1.22 to 3.67 x10–3 L µg–1, respectively. Beek et al. (1980) and Van der Zee and Van Riemsdijk (1986) suggested that {Gamma}{infty} for soil P sorption can be expressed as a function of oxalate Fe + Al. They pointed out that oxalate-extractable Fe and Al in soil exist primarily as reactive amorphous (hydr)oxides that can effectively bind P. However, we did not find such a relationship (Table 7), and there was almost no correlation between oxalate P and oxalate Fe + Al (r2 = 0.02, p = 0.47). Considering that P levels overall were high (Table 2), the association between P and amorphous Fe and Al (hydr)oxides is not strong enough to be considered the primary P retention mechanism in MRB soils.


View this table:
[in this window]
[in a new window]
  		Table 7. Predictions of P sorption parameters from linear regressions on soil properties (n = 25).

 
Although the oxalate Fe + Al content of a soil could not be used to predict {Gamma}{infty} for all MRB soils, it did correlate well with the oxalate-extractable P of the eight acidic (pH < 6.0) MRB soils (r2 = 0.93, p < 0.001; Fig. 2) . The slope of the equation is similar to the average ratio (0.29) between oxalate P and oxalate Fe + Al obtained by van der Zee et al. (1987) for 24 acidic soils with a wide range of oxalate Fe + Al content in the Netherlands. Furthermore, the correlation between {Gamma}{infty} and oxalate Fe + Al for the eight acidic soils was much higher (r2 = 0.44, p = 0.07), compared with that for all 25 soils (r2 = 0.01, p = 0.62; Table 7). Thus, it appears that in acidic soils of the MRB, amorphous Fe and Al are important P retention agents. This is consistent with the generally accepted understanding that the formation of Fe and Al phosphates at low pH drives P adsorption in acidic soils.



View larger version (15K):
[in this window]
[in a new window]
  		Fig. 2. Correlation between oxalate-extractable P (Oxalate-P) and oxalate-extractable Fe and Al (Oxalate-[Fe+Al]) for eight acidic soils of the Minnesota River Basin.

 
Among the STPs, TP had the best correlation with {Gamma}{infty} (r2 = 0.72, p < 0.001, Table 7). This makes intuitive sense because soils with high {Gamma}{infty} are expected to retain more P, and the long history of P fertilizer usage in the MRB made this P retention and accumulation possible. The other STPs also had positive correlations with {Gamma}{infty} at p < 0.001. Novotny and Chesters (1978) correlated {Gamma}{infty} with the clay and organic content of acidic soils:

[7]
We used these variables to predict {Gamma}{infty} for MRB soils and found that {Gamma}{infty} was poorly predicted by these characteristics individually or in combination (Table 7). Although correlations with pH and CCE were statistically significant (p < 0.05), the r2 values are too low to yield reliable predictions. Regression of {Gamma}{infty} against organic matter content and pH also did not yield a close fit.

Phosphorus Sorption Energy Constant (b)
The sorption energy constant (b) was correlated more strongly with soil properties than {Gamma}{infty} was (Table 7). Organic content, pH, and CCE all showed significant correlation with b. Linear regression of b against pH, clay content, and organic content yielded the best fit:

[8]
A regression using hydrogen ion activity (10–pH) yielded an R2 of only 0.38 (n = 25, p = 0.02, Table 7). In contrast, the best-fit equation found by Novotny and Chesters (1978) was:

[9]
The sign of the coefficient for organic matter content is different in the two equations. For MRB soils, P sorption energy decreased when the percentage of OM increased; the opposite was true for the soils used for Eq. [9]. This difference cannot be explained by the fact that Eq. [9] was based only on acidic soils (pH < 7.0) while Eq. [8] was based on both acidic and calcareous soils; a regression equation for the nine acidic soils (pH < 6.0) with calculated b also had a negative sign for the coefficient of percentage of OM (although not significant at p < 0.05; equation not shown). Soil organic matter is negatively charged even in the acidic pH range (Sposito, 1989). In the pH range of our soils (4.9–8.1) and for our experiments, P existed mainly in the forms of H2PO–4 and HPO2–4. Consequently, if the organic matter in the soils was associated with soil particles, P adsorption onto the particles could be hampered by negatively charged OM. The generally high organic content of the MRB soils (Table 2) implies that the association between soil OM and soil particles could be significant. In addition, increasing soil organic content could increase the number of mineral P sorption sites blocked by organic matter. As a result, sorption energy decreased as the soil OM content increased.

The Langmuir b was negatively correlated with total P of the soils (Table 7). All other STPs except NaOH P also had negative correlations with b (p < 0.05), but the opposite was true for {Gamma}{infty} (all p < 0.01). This contrast indicates that as more P was adsorbed, the binding between adsorbed P and soil particles became weaker. The Langmuir model assumes a uniform sorption energy of the sorbate on the sorbent. The sorption energy constant obtained from the Langmuir model thus is an indication of the average energy over all sites.

Phosphorus sorption has been described as a two-phase process—a rapid initial step followed by a long, slower step—for goethite (Atkinson et al., 1972) gibbsite (Kyle et al., 1975), kaolinite (Chen et al., 1973), soils (e.g., Munns and Fox, 1976), and lake sediment (Detenbeck and Brezonik, 1991). The slower step has been explained by following mechanisms: (i) fixation of P into crystalline phases (Chen et al., 1973; Munns and Fox, 1976); (ii) diffusion of P into the interior of particles (Barrow, 1983); and (iii) existence of different populations of sites responsible for the sorption (Syers et al., 1973). The sorption energies for the first two slow-step mechanisms are presumably higher than that for the rapid step. For the third mechanism, sorption energy is assumed to vary among populations of sites. Within the time frame of our experiments (24 h), it is unlikely that the first two slow mechanisms were important. Consequently, the decrease in b with increasing sorption of P suggests that there was more than one population of sorption sites on the soils and that sites with higher sorption energy became occupied earlier. In addition, {Gamma}{infty} and b were negatively correlated r2 = –0.58, p < 0.01, n = 25), suggesting that most of the increased P sorption capacity was accounted for by sites with lower binding energy.

Phosphorus Sorption Saturation
Both PSIm and PSIs were positively correlated (Table 8) with the seven STPs and with two soil sorption capacity indicators, EPC, and desorbability (all at p < 0.001 except for PSIm versus TP). Because NaOH P is part of PSIm and PSIs (Eq. [3] and [4]), correlations between NaOH P and the index values may be spurious. Correlation coefficients for PSIs with the other eight variables—except desorbability are higher (or of equal value) compared with correlation coefficients of PSIm with the corresponding variables. As pointed out earlier, the ability of a soil to retain P depends on both the amount it can adsorb ({Gamma}{infty}) and the strength of binding (constant b). Including b in the PSIs (Eq. [4]) improved the ability to describe soil characteristics related to P retention. In a related study, we found that PSIs was better than PSIm in predicting the ABP concentration in simulated runoff from a subset of the same soils (Fang et al., 2002). In that study ABP (approximated with Fe-paper extraction) in simulated runoff was predicted from PSIm and PSIs. In the current study, the direct correlation between soil ABP (measured by bioassay) and these two soil P saturation indices was examined.


View this table:
[in this window]
[in a new window]
  		Table 8. Coefficients of determination (r2) between P sorption saturation indices and nine soil characteristics; n = 25; all significant at 0.001 except as noted by * (significant at 0.05).

 
The fact that desorbability had a higher correlation with PSIm rather than PSIs (Table 8) likely is due to the conditions under which desorbability was determined (24 h of aqueous extraction immediately after a 24-h sorption period). In such a short reaction period (compared with field conditions), strong bonds between P and soil particles are not likely to be established (Chen et al., 1973; Munns and Fox, 1976; Barrow, 1983). Therefore, sorption capacity may be more important than sorption energy in determining the amount of P that will be desorbed, and PSIm, which takes into account only the sorption maximum, correlated better with desorbability than PSIs, which includes binding energy.

To apply the PSIs concept to quantifying soil ABP in the MRB, we regressed chl a against PSIs (Fig. 3) . Compared with other predictors (Table 3), PSIs gave the best estimate (r2 = 0.72) of chl a response and thus is the most reliable indicator of the portion of soil P related to water quality in the MRB. In our simulated runoff study (Fang et al., 2002), we concluded that PSIs values of 20 to 26 could cause eutrophication in receiving streams. Based on the relationship between PSIs and chl a response (Fig. 3), this PSIs range translates into a critical soil chl a response range of 6.4 to 6.9 µg L–1. Of the 23 soil samples with measured chl a response in this study, 10 were above this range and 11 were below.



View larger version (15K):
[in this window]
[in a new window]
  		Fig. 3. Correlation between chlorophyll a (chl a) response and P saturation index based on sorptivity (PSIs) (n = 23).

 
Based on the relationship between PSIs and Bray P (Fig. 4) , the critical PSIs range translates into a critical Bray P range of 39 to 57 mg kg–1, which is similar to the range (35–50 mg kg–1) calculated from the relationship between Bray P and runoff ABP in our simulated runoff study. Bray P is the most common soil-P test used in the MRB, and Bray-P data are widely available for the basin. However, both Mehlich-III P and Fe-paper P have better correlations with PSIs (Table 8) and chl a response, and better estimates of ABP thus could be obtained from these STPs if such data were available. We calculated critical values of these STPs from their relationships with chl a. For Mehlich-III P, the critical range is 62 to 82 mg kg–1 (based on the regression relationship: chl a response = 0.0228 x Mehlich-III P + 5.0186; r2 = 0.65, p < 0.001). For Fe-paper P, the critical range is 22 to 27 mg kg–1 (based on Eq. [5] above). Use of these STPs and critical values for defining excessive amounts of soil ABP should be promoted in the MRB.



View larger version (15K):
[in this window]
[in a new window]
  		Fig. 4. Correlation between P saturation index based on sorptivity (PSIs) and Bray P (n = 25).

 

   CONCLUSIONS
TOP
 ABSTRACT
 INTRODUCTION
 MATERIALS AND METHODS
 RESULTS AND DISCUSSION
 CONCLUSIONS
REFERENCES
 
For soils of the MRB, ABP (represented by chl a bioassay response) was correlated (p < 0.001) with Bray P, Mehlich-III P, NaOH P, oxalate P, and Fe-paper P. Among them, Fe-paper P predicted ABP best, particularly for calcareous soils. Among the STPs, Fe-paper P was most strongly correlated with Mehlich-III P for both acidic and calcareous soils. For acidic soils, amorphous Fe and Al apparently were the primary P retention agents in soil particles.

Soil sorption data were well described by the Langmuir model. The Langmuir sorption maximum ({Gamma}{infty}) was not correlated with basic soil physical and chemical properties such as clay content and OM content, but the sorption energy constant (b) could be predicted from a regression equation using soil pH, clay content, and organic content as independent variables. This equation also showed that b was inversely correlated to soil organic content, suggesting that OM plays a negative role in soil P sorption. In addition, sorption energy decreased as more P was adsorbed on soil particles, suggesting that more than one population of sorption sites with different sorption energy existed on the surface of soil particles.

The P saturation index with sorptivity ({Gamma}{infty} x b) as the measure of sorption capacity PSIs, gave the best estimate of soil ABP among all predictors used in this study. The PSIs itself can be estimated from Fe-paper P, Mehlich-III P, and Bray P. Quick estimation of soil ABP thus is possible in the MRB where STP data are widely available. Ranges of the three STPs and PSIs indicative of critical values of soil ABP were derived from their relationships with chl a response: 22 to 27 mg kg–1 for Fe-paper P; 62 to 82 mg kg–1 for Mehlich-III P; 38 to 54 mg kg–1 for Bray P; and 20 to 26 for PSIs.


   ACKNOWLEDGMENTS
 
This study was sponsored by an EPA-funded 319 grant from the Minnesota Pollution Control Agency and an EPA-NSF water and watersheds research project (EPA No. 825290010). The authors thank the late Bill Nord from the Research Analytical Laboratory at the University of Minnesota, St. Paul for his assistance and excellent work in soil chemical analyses and three anonymous reviewers for their helpful criticism of this manuscript.

Received for publication April 1, 2003.


   REFERENCES
TOP
 ABSTRACT
 INTRODUCTION
 MATERIALS AND METHODS
 RESULTS AND DISCUSSION
 CONCLUSIONS
 REFERENCES
 




This article has been cited by other articles:


Home page
 Soil Sci.Home page
C. H. Bolster and G. M. Hornberger
On the Use of Linearized Langmuir Equations
Soil Sci. Soc. Am. J., September 28, 2007; 71(6): 1796 - 1806.
[Abstract] [Full Text] [PDF]

This Article
Right arrow Abstract Freely available
Right arrow Figures Only
Right arrow Full Text (PDF) Free
Right arrow Alert me when this article is cited
Right arrow Alert me if a correction is posted
Services
Right arrow Similar articles in this journal
Right arrow Similar articles in ISI Web of Science
Right arrow Alert me to new issues of the journal
Right arrow Download to citation manager
Right arrow reprints & permissions
Citing Articles
Right arrow Citing Articles via HighWire
Right arrow Citing Articles via Google Scholar
Google Scholar
Right arrow Articles by Fang, F.
Right arrow Articles by Hatch, L. K.
Right arrow Search for Related Content
PubMed
Right arrow Articles by Fang, F.
Right arrow Articles by Hatch, L. K.
GeoRef
Right arrow GeoRef Citation
Agricola
Right arrow Articles by Fang, F.
Right arrow Articles by Hatch, L. K.
Related Collections
Right arrow Water Quality
Right arrow Nutrient Management
Right arrow Soil Chemistry

HOME HELP FEEDBACK SUBSCRIPTIONS ARCHIVE SEARCH TABLE OF CONTENTS
The SCI Journals Agronomy Journal